Appendix 1

Single-Degree-of-Freedom
Dynamic Systems with Damping

A.l VISCOUS DAMPING

This brief description attempts to illustrate differences and special features of

some of the various damping mechanisms typical for mechanical systems. A
classic single-degree-of-freedom (SDOF) mechanical system in Fig. A. 1.1 com-
prises mass rn, spring k, and viscous damper c. The equation of motion of this
system at free vibration condition (no external forces, Fo = 0, uf = 0) can be
written as

my + cy + ky = 0 (A. 1.la)

where y = displacement of mass rn. When viscous friction in the damper is not
very intense, c < 2 dkm,then the solution of (A.l.1) is

y = e-"(C, sin o*t + C2cos o*t) (A. 1.2)

where

n = -C o* = 40; - n2 (A. 1.3)

2m'

where oo= natural frequency of the system without damping (c = 0), and con-
stants CI and C2 are determined from initial conditions y ( 0 ) = yo, y = yo as
398

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

SDOF Systems with Damping 399

Figure A.l.l Classic single-degree-of-freedom (SDOF) mechanical system.

C1 = ( y o + nyo)/o*; C2 = y o (A. 1.4)

Another format of the solution is

y = e-n‘ sin(o*t + p) (A. 1.5)

where

(A. 1.6)

It can be seen from Eqs. (A.1.2) and (A.1.5) that variation of y in time (motion)
is a decaying oscillation with a constant frequency (I)* and gradually declining
amplitude (Fig. A.1.2). Parameter j3 is called the loss angle and tan p is the loss
factor.

Figure A.1.2 Decaying oscillation with a constant frequency o* and gradually declin-
ing amplitude.

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

400 Appendix 1

Envelopes of the decaying time history of y are described by functions

A = t A o e '" (A.1.7)

where A. = ordinate of the envelope curve at t = 0.

The ratio of two consecutive peaks A ( t ) : A ( t+ T*), which are separated by
time interval T* = 2n;/o* (period of the vibratory process), is e"" = constant.
The natural logarithm of this ratio is called logarithmic (or log) decrement. It is
equal to

(A.1.8)

Thus for the same damper (damping coeflcient c), log decrement of a system
with viscous damper depends on stiffness k and mass m of the system.
For forced vibrations (excitation by a harmonic force F = Fo sin cot applied
to mass m ) , the equation of motion becomes

(A.l. lb)

and the amplitude of response is

Expression (A.1.9a) is plotted in Fig. A.1.3 for various 6.

Frequently, there is a need for vibration isolation. Two basic cases of vibra-
tion isolation are: (1) protection of foundation from force F = Fo sin cot generated
within the object (machine) represented by mass m ; and (2) protection of a vibra-
tion-sensitive object (machine) represented by mass m from vibratory displace-
ment of the foundation a = u r sin cot.
In the first case, the force transmitted to the foundation is F, sin Ci)t and the
quality of 1 ibration isolation is characterized by force transmissibility TF = F,/
Fo. In the second case, the displacement transmitted to mass m is characterized
by displacement transmissibility T,, = af,,/u+. For the single-degree-of-freedom
isolation system, which can be modeled by Fig. A. 1.1

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

SDOF Systems with Damping 401

A
Fdk

4.0

3.0

2.0

1.0

Figure A.1.3 Hysteresis loop illustrating change of deformation y for the processes of
increasing (loading) and decreasing (unloading) force P.

(A.1.9b)

*I):( I:():(
+ +

) plotted by solid lines in Fig. A. 1.4 for several values of 6.

Expression (A. 1 . 9 ~is
It is important to note that while higher damping results in decreasing transmissi-
bility T around the resonance (o= o,), it also leads to a significant deterioration
of isolation (increased T ) above o = 1.41 oo(isolation range).

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

402 Appendix 1

T
8G
6G

40
JG

ZG

I0
8
6
4
3

7
08
06

04
a3

d2

Figure A.1.4 Transmissibility of single-degree-of-freedom vibration isolation system

in Fig. A . l . l vs. frequency and damping. Solid lines indicate viscous damping; broken
lines indicate hysteresis damping.

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

SDOF Systems with Damping 403

A.2 -
HYSTERESIS I NDUCED DAMPI NG

Deformation of mechanical components and joints between them (contact defor-

mation) is not perfectly elastic. This means that deformation values during the
process of increasing the external force (loading) and decreasing the external
force are not the same for the same magnitudes of the external force. This effect
results in developing of a hysteresis loop (Fig. A.1.5), which illustrates change
of deformation y for the processes of increasing (loading) and decreasing (un-
loading) force P. The area of the hysteresis loop represents energy lost during
one loadinghnloading cycle. It is established by numerous tests that for the ma-
jority of structural materials, as well as for joints between components, the area
of the hysteresis loop does not strongly depend on the rate of force change (i.e.,
on frequency of the loading process), but may depend on amplitude of load/
deformation. This statement can be formalized as an expression for the energy
Y lost in one cycle of deformation as function of amplitude of deformation A

where a and r = constants.

To derive the law describing the decaying oscillatory process when the
damping is due to hysteresis, the loss of energy during one cycle can be equated
to change of system energy during one cycle of vibration. Let's consider one
period of the process in Fig. A. 1.2 and start the period when the displacement
peaks at amplitude A(t). At this moment the kinetic energy of the mass is K =
0, and all energy is stored in the spring k (Fig. A. 1.1) as the potential energy V.
In the beginning of the period

Figure A.1.5 Hysteresis loops associated with a deformation process.

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

404 Appendix 1

v,, = "kA'(t) (A.1.11)

2

At the end of the period

1
vp = -kAyt + T") (A. 1.12)
2

The increment of the potential energy is

AV = V p - V,, = '/2 k[A'(t + T " ) - A'(!)] (A. 1.13a)

= '/? k [ A ( t + T") + A ( t ) ] [ A ( t+ T * ) - A(t)l

The sum inside the first square brackets on the right-hand side is -2A(t) = 2A
if the energy loss during one cycle is not very large. The difference inside the
second bracket is -AA, and

-AV = kAAA (A. 1.13b)

This increment or the potential energy is equal to the energy loss (A.l.lO), or

(A.1.14a)
or
(A. I. 14b)

This expression defines the shape of the upper envelope of the oscillatory process.
Considering this envelope as a continuous curve A = A(t), approximately

AA = T * ( d A / d t )= ( 2 n / ~ * ) ( d A / d t ) (A.1.15)

From (A.1.15) and (A. 1.14b)

dAldt = - ( aW" / 2 n k)A ' (A. 1.16)

If Y = 1, the solution of Eq. (A.l.16) is an exponential function

The ratio of two peak displacements A(t)/A(t + T * ) = ea" is again constant, as

in the case of viscous damper, but in this case the log decrement does not depend
on mass rn

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

SDOF Systems with Damping 405

6 = a/k (A.1.18)

Although for Y = 1 the log decrement does not depend on amplitude of vibration,
for many real-life materials and structures value of Y may deviate from 1, and
then the log decrement would be changing in time with the changing vibration
amplitude. The character of the change is illustrated in Fig. A. 1.6 by envelopes
for the vibratory process at Y = 0 and Y = 2, in comparison with the exponential
curve (amplitude-independent log decrement) for Y = 1. Dependence of log dec-
rement on amplitude for fibrous and elastomeric materials is illustrated in Fig.
3.2. For many elastomeric materials (rubber blends) Y = -1.
The response amplitude for the forced vibrations for a system with hysteresis
damping

Fo
A = (A.1.19a)

For Y = 0 (Coulomb friction-induced damping), for Y = 2, and for other Y + 1, the

amplitude of the forced vibrations can be found after solving equation (A. 1.19).
However, for Y = 1, which is typical for many rubber blends,

(A.1.19b)

k j(1 - $)2 + ($2 j( $)*+ (!)*

1 -

Equation (A.1.19b) is similar to the expression for the response to external har-
monic excitation by a system with viscous friction described by Eq. (A.1.9a),

Figure A.1.6 Amplitude dependencies of log decrement 6 for different damping mecha-
nisms.

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

406 Appendix I

with the only difference being the second (damping) term under the radical sign.
For a system with viscous friction this term is frequency dependent, while for a
system with hysteresis damping it does not depend on frequency.
Transmissibility T between mass m and foundation for the force excitation
of the mass F = Fo sin ot or between foundation and mass rn for the “kinematic”
excitation of the foundation a = sin ot for the system with the hysteretic
damping when r = 2 is

(A. 1 . 1 9 ~ )

The differences between (A.1.19~)and (A.1.9b) are due to presence in the

latter and absence in the former of the frequency ratio multipliers in the second
terms under the radical signs both in the numerator and in the denominator. Effect
of these seemingly subtle differences is practically indistinguishable in the fre-
quency range o = 0 - 1.41 oo, as well as for small damping (6) values. However,
for larger damping values the difference in the isolation effectiveness (at o >
1.41 coo) is very pronounced, as shown by broken lines in Fig. A.1.4. Although
vibration isolators having significant viscous damping demonstrate deteriorating
performance in the “isolation range,” 03 > 1.41 oo, vibration isolators with the
hysteretic damping characterized by r = 1 provide effective performance in the
“isolation range’ ’ while limiting transmissibility around the resonance frequency.
A system with hysteresis damping with the hysteresis damping can be de-
scribed by Eqs. (A. 1.1a) and (A. 1.1b) if the damping coefficient c is replaced by
a frequency- and amplitude-dependent coefficient

L‘/? = aA’-’/no (A.1.20a)

which for r = 1 becomes only frequency-dependent

(‘/] = a/no (A. 1.20b)

The loss factor for a system with hysteresis damping is

(A.1.21)

A.3 IMPACT D A M P I N G [I]

Impact interactions between mechanical components cause energy losses. Such

impact interactions can occur in clearances between components in cylindrical

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

SDOF Systems with Damping 407

joints or guideways, or in specially designed impact dampers. During a typical

vibratory process in a system with clearances, impacts occur every time the sys-
tem passes its equilibrium configuration. The amount of energy loss during im-
pact can be presented as a function of relative velocity of the coimpacting compo-
nents between the impact moment

Y = bv2 (A. 1.22)

where b = a constant having dimension of mass.

Let's consider a half-period of oscillation that commences at a maximum
displacement A(0). In the first quarter of the period the mass is moving with a
constant energy '/2M2(0)and the square of velocity at the end of this quarter of
the period is

v 2 = (k/m)A2(0) (A. 1.23)

At this moment an impact occurs and, consequently, a loss of energy by the

amount of Eq. (A.1.22). After the impact, the system moves with energy

kA2(0)/2- (bk/m)A2(0)= '/2kA2(0)(1- 2b/m) (A. 1.24)

This energy remains constant during the second quarter of the period. Accord-
ingly, at the end of the second quarter the potential energy is equal

V2kA2(T/2)= V2kA2(0)(1- 2b/m) (A. 1.25)

Thus, the ratio of maximum displacements is

(A. 1.26)
A(;) d q
The same ratio would materialize for the next half-period. Thus, for the
whole period

A(O)/A(T)= 1 / ( 1 - 2b/m) = constant (A. 1.27)

Since the ratio of sequential maximum displacement is constant, the envelope of

the time plot of the decaying vibratory process is an exponential curve

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

408 Appendix 1

A = AOep’lr (A.1.28)

This process is associated with the log decrement

6 = nt = log,.[l/(l - 2b/rn)] = -log,(l + 2blrn) = -2blnz (A.1.29)

if 2blm is small.

REFERENCE

1. Panovko, Ya.G., Introduction to Theory of Mechanical Vibration, Nauka Publishing

House, Moscow, 1971 [in Russian].

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

Appendix 2
Static Stiffness Breakdown for
Cylindrical (OD) Grinders

Although modal analysis of complex structures is very useful for determining

weak links in complex mechanical structures, it can be complemented by measur-
ing static deformations of various components and their connections under loads
simulating working conditions of the system. Evaluation of the static compliance
breakdowns allows us to better understand the role of small but important compo-
nents, more readily simulate diverse working conditions/regimes, detect and
study nonlinear deformation characteristics of some components that can confuse
the modal analysis procedure, and more. Since measuring the compliance break-
down under static loading is much more time-consuming than the dynamic evalu-
ation, the former should be undertaken in cases of critical importance or when
the nonlinear behavior is strongly suspected. This appendix is a summary of a
detailed analysis of static deformations for cylindrical OD grinders for grinding
parts up to 140 mm in diameter and up to 500 mm long [I]. However, this study
is rather generic, and its techniques and conclusions can be very useful for other
mechanical systems.
The machine is sketched in Fig. A.2.1. The most stiffness-critical units are
identified as follows. Wheelhead 3 houses spindle 1 supported in hydrodynamic
bearings 2 and is mounted on a carriage comprising upper 4 and lower 8 housings.
Hydraulic cylinder 6 and ball screw 15 with nut 16 are mounted in housing 8.
The machined part is supported by headstock 21 and by tailstock (not shown in
Fig. A.2.1) installed on angular (upper) table 19, which is attached to longitudinal
(lower) table 18 moving in guideways along bed 11.
One end of ball screw 15 is supported by bracket 17 fastened to housing 4;
the other end is driving pusher 13. The force between pusher 13 and ball screw
15 is adjusted by compression spring 12 and threaded plug 10, which are housed

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

410 Appendix 2

Figure A.2.1 Test forces P , , P 2 ,and P3 and measurement positions/deformation sensors

I-IX for cylindrical (OD) grinder.

in bracket 9 fastened to housing 4. Thus, spring 12 generates preload (up to 1,000

N) in connection: bracket 17-ballscrew 15.
Wheelhead 3 may perform setup motions along guideways on housing 4,
with its final position secured by set screw 5 ; fast motion (before and after ma-
chining) by piston 7 moving in hydraulic cylinder 6; and feed motion by ballscrew
15 driven by worm gear 14. During the feed motion, piston 7 is touching the left
face of cylinder 6.
The principal contributors to stiffness/compliance breakdown are: spindle 1
in bearings 2; joint between wheelhead 3 and housing 6; joint between bracket
17 and screw 15; connection between ball screw 15 and ball nut 16;joint between
table 18 and guideways; headstock 21 (or tailstock); and joint between supporting
center 23 and housing of headstock or tailstock.
In some similar grinders the wheelhead can be installed directly on rolling
friction guideways of the lower housing of the carriage, not on the upper housing
as shown in Fig. A.2.1. This does not change the compliance breakdown; it is
shown below that displacement of the wheelhead relative to the upper housing
is only about 0.5 pm under 600 N load.
Test forces were applied (through load cells) in several locations (Fig. A.2.1):
between nonrotating wheel and part ( P , ) ;between bracket 22 attached to table
19 and wheel head (P?);and to end face of screw 15 (Pi). The forces were varied
in two ranges 0-300 N and 0-600 N. Displacement transducers were located in
positions I-IX: I and I1 measured displacement of wheel head 3 relative to table

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

Breakdown for Cylindrical Grinders 41 1

19 and bed 11; I11 measured displacement of housing 4 relative to bed 11; IV
measured the joint between end face of nut 16 and housing 8; V measured the
contact between end of screw 15 and housing 8; VI and VII measured displace-
ments of tables 18 and 19 relative to bed 11; VIII measured displacement of
tailstock (headstock) relative to bed 11; and IX measured displacement of sup-
porting center relative to table 19.
Figure A.2.2 shows measuring setups for displacements of the part as well
as of the headstock and tailstock and their components. Transducers X-XI11 are
on the tailstock side and measure displacements of part 3, supporting center 4,
holder 5 , mounted in sleeve 7 on balls 6, and tailstock 8 relative to upper table
9. Transducers XIV-XVII are on the headstock and measure displacement of
part 3, supporting center 2, central bushing 1, and headstock 10 relative to table
9, respectively.

! ? 4 5 6 7 8

10
.- a Y

Figure A.2.2 Part support schematics used for measuring deformations of various com-
ponents under radial load P;X-XVII measurement positions/deformation sensors.

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

412 Appendix 2

Stiffness of the part as well as of components of headstock and tailstock

were determined under force P applied to an end of the part from the headstock
or tailstock (12 and 11 are, respectively, arm lengths of load application). To
determine influence of stiffness of various components of headstock and tailstock
on the total part stiffness relative to table 9, measurements were performed not
only in setup of Fig. A.2.2a, but also as shown in Figs. A.2.2b, c, and d. For
example, to identify influence of supporting centers stiffness relative to table 9
on the part displacements, bracket 11 or 12 was attached to front or rear center,
respectively, and the load was applied to the part. Displacements of ends of these
brackets were measured at distances I , and l2 from the head-/tailstock. Influence
of stiffness of center holder 5 relative to the table on effective stiffness of the
part was measured in a similar way. To perform this measurement, bracket 13
was attached to the holder (Fig. A . 2 . 2 ~ )and displacements of the bracket end
were recorded.
Supporting centers 2 and 4 have Morse taper #4. Depending on grinding
conditions the centers could be short or long (the cylindrical part is 20 mm
longer). Overhang of the center holder of the tailstock was varied during grinding
within 20 mm. Distance I , = 85 mm for the short center and minimum holder
overhang; 105 mm for the short center and the maximum overhang as well as
for the long center and minimum overhang maximum overhang; and 125 mm
for the long center and maximum overhang. Bushing 1 is fixed stationary in the
headstock; thus I? depends only on length of center 2: l2 = 160 mm for the short
center and 180 mm for the long center.
Plots in Fig. A.2.3 show displacements 6 vs. radial load for wheel head,
tailstock and headstock, and other components. Spindle stiffness (245 N/pm)
was determined from computed stiffness of its hydrodynamic bearings since the
measurements were performed without spindle rotation. The broken line in Fig.
A.2.3a is plotted by adding computed spindle displacements to the measured
(transducer 11) displacement of the wheelhead. This line represents the total effec-
tive compliance of the spindle due to compliance of all components sensing the
forces from the wheelhead. These plots show that the largest contributor to com-
pliance of the wheelhead is the ball screwhut transmission (the distance between
lines IV and V). The next contributor is spindle (the distance between the broken
line and line II), the third is joint screw 15-bracket 17 (the distance between
lines V and 111). Other displacements/compliances are very small and can be
neglected.
Stiffness of the upper and lower tables is different when measured at the
tailstock and at the headstock (Fig. A.2.3b and c). This is mostly due to manufac-
turing imperfections of the guideways resulting in their nonuniform fit along the
length and thus in stiffness variations. Since the radial force is applied at the
centerline level, it also generates angular displacements of the lower table in

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

Breakdown for Cylindrical Grinders 41 3

3: Mm a

Figure A.2.3 Deformations 6 (pm) of components of (a) wheelhead, (b) tailstock, and
(c) headstock relative to bed under radial force P at different measuring points. The dashed
line is the calculated displacement of spindle.

the guideways in the transverse vertical and horizontal planes. As a result, dis-
placements of tables as well as the tailstock and headstock become uneven.
Stiffness of the headstock and tailstock depends on the attachment method
of each unit to the table. The headstock is fastened by two short bolts 20 (Fig.
A.2.1). The tailstock is fastened by one long bolt in the middle of its housing.
This arrangment simplifies resetting of the tailstock bit reduces its stiffness. It
can be concluded from lines VIII in Fig. A.2.3b and I1 in Fig. A.2.3a that compli-
ance of the tailstock (taking into account also deformations of lower and upper
tables) is close to compliance magnitude of the wheelhead. Figure A.2.4 gives
deformations of components of both headstock and tailstock for various combina-
tions of lengths of the centers and overhang of the sleeve. Lines X-XI11 in Figs.
A.2.4a-d represent deformations of the machined part, supporting center, sleeve,
and tailstock, relative to the upper table measured in positions indicated in Figs.
A.2.2a-d, respectively. Lines VIII in Fig. A.2.4a-d show the total displacement
of the part in relation to the bed; lines I1 show the part in relation to the wheelhead;
and the broken lines show the part in relation to the wheel spindle. Lines XIV-
XVII in Figs. A.2.4d-e show displacement of the part, the supporting center
sleeve, and headstock, respectively, in relation to the upper table.
These plots give an understanding of influence of deformations of each main
structural component of the machine tool on the spindle-part deformations under
forces up to 300 N. The data in Table 1 shows that the most compliant elements
of the tailstock are the supporting center and the sleeve. Depending on the length

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

 Appendix 2

S./cl.m P

0 60 120 180 240P,N 0 60 120 180 2401jN

e f
Figure A.2.4 Displacements 6 (pm) of components of (a-d) tailstock and (e-f) head-
stock at: (a) short support center with minimal sleeve overhang; (b) short center with
maximum overhang; (c) long center with minimal overhang; (d) long center with maxi-
mum overhang; (e) short center; and (f) long center. The dashed line is spindle displace-
ment.

Table A.2.1 Compliance Breakdown of OD Grinder

Spindle Contribution of unit deformation (%)
Setup of displacement
Fig. A.2.4 (%)
~~~
I1 VTII x XI XI1 XI11 XIV xv XVI XVII

a 7.7 14.2 17.4 3.2 21.7 26.0 9.8 - - - -

b 6.8 12.5 15.5 4.8 21.7 29.0 9.7 - - - -

C 4.6 9.4 11.2 5.2 39.7 18.8 11.1 - - - -

d 4.4 9.1 10.7 6.0 38.9 18.8 12.1 - - - -

e 9.6 20.1 5.2 - - - - 11.1 32.8 2.5 18.7
f 6.1 12.7 3.3 - - - - 8.5 54.3 1.4 13.7

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

Breakdown for Cylindrical Grinders 415

of the center, it is responsible for 21.7-39.7% of the total displacement of the

spindle, and the sleeve is responsible for 18.8-29%.
The tailstock displacement in relation to the upper table (measuring position
XIII) and in relation to the bed (position VIII), as well as displacement of the
wheelhead in relation to the bed have close magnitudes. The most significant
component of the headstock deformation is the supporting center (the short center
accounts for 32.8%, the long center for 54.3% of the total deformation of the
tailstock spindle). The spindle displacement accounts for 29.8% of the total dis-
placement with the short center and 18.8% with the long center.
The plots in Fig. A.2.4 indicate that stiffness of the same machine tool (not
considering deformations of the machined part) varies, for loading forces to 300
N, from 16.3 to 10.1 N/pm at the headstock.
Thus, the most significant component in the compliance breakdown is com-
pliance of the supporting centers. Displacement of the loaded center is

6 = 6, + 6, + 6 , (A.2.1)

where 6b = bending deformation; 6, = radial displacement due to contact defor-

mations; and 6, = displacement at the load application point due to angular con-
tact deformations. From Marcinkyavichus and Yu [ I]

6, = P(Ako.s + Bk"")); 6, = P(Cko.2s+ Dk'.')); (A.2.2)

where P = radial force; A, B, C, and D = constants depending on the Young's

modulus of the center and the sleeve, taper diameter, and distance from the load
application point to the sleeve face; k = parameter depending on quality of the
tapered connection sleeve = 0.1-1.O pm-mm2/N.
For cases illustrated in Fig. A.2.4 at the force 300 N, 6, = 1.6 pm for the
short center and 6, = 4.2 pm for the long center. Separate tests performed on
the centers gave 6, + 6, = 3-4 pm for the short center and 5.8-6.8 pm for the
long center, and k = 0.4-0.8 pm-mm2/N. Additional tests on other machine
tools and on other supporting centers resulted in values k = 0.4-1.2 pm-mm2/
N. Displacements of the wheelhead relative to the upper table (measuring position
I) under the force from bracket 22 to wheel head 3 (Fig. A.2.1) are close to
displacement of the wheelhead relative to the bed (line I1 in Fig. A.2.3a) and to
displacement of the tailstock relative to the bed (line VIII in Fig. A.2.3b).
To enhance the total stiffness of the grinder, the most important are stiffness
values of the supporting centers, the sleeve, and the tailstock, as can be seen
from Fig. A.2.4. Reducing values of parameter k can enhance stiffness of the
centers. This can be achieved by improving the fit in the Morse taper connection
between the center and the sleeve or by preloading this connection. The contact
area in the tapered connections of the grinder on which the tests were performed

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

416 Appendix 2

was about 70-80% of the nominal contact area; it was not possible to obtain k
< 0.4 pm-mm’/N. Mutual lapping of the tapered connections was out of consid-
eration since the centers have to be frequently replaced depending on the grinding
conditions. Axial preloading of the tapered connection resulted in bulging of the
sleeve (as in Fig. 4.32) and in over preloading of the balls guiding the sleeve
motion. To maintain accessibility of the grinding wheel to the part, the cylindrical
part and the 60” “center part” of the supporting center can be made like in the
standard Morse taper #4 while the tapered seat should be dimensioned as the
standard Morse taper #5. Such arrangement would reduce 6, + 6 , up to 2.5 times.
An analysis using data on contact deformations (see Chapter 4) has shown
that 20-30% (depending on the overhang) of the sleeve deformation is due to
compliance of the balls guiding the sleeve in the holder. The other 80-70% is
due to bending of the sleeve inside the ball bushing. Reduction of the bending
deformation can be achieved by increasing diameter of the sleeve or the number
of guiding balls, or by reduction of the overhang. The first approach is unaccept-
able since in case of grinding of a tapered part between the centers with angular
displacement of the upper table in relation to the lower table, the tailstock would
interfere with the wheelhead. The third approach cannot be realized since the
overhang in the present design is the minimum acceptable one. Thus, some reduc-
tion of deformations and the resulting stiffness enhancement can be achieved by
increasing the number of balls and by their optimal packaging. Another approach,
using rollers instead of balls, would enhance stiffness but would result in a sig-
nificantly more complex and costly system.
Stiffness of the tailstock could be enhanced by using stiffer attachment to
the table, e.g., by using two bolts instead of one. However, it would lengthen
the setup time, which might not be desirable.

REFERENCE

1. Marcinkyavichus, A.-G. Yu, “Study of stiffness of cylindrical OD grinders,” Stanki

i instrument, 1991, No. 2, pp. 2-4 [in Russian].

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

Appendix 3
Influence of Axial Force on Beam
Vibrations

The deflection equation of the beam in Fig. 7.30a subjected to a distributed load
p can be written as [l]

(A.3.1)

If the distributed load p is due to inertia forces of the vibrating beam itself, then

a2Y
p = - m T (A.3.2)
at

where rn = mass of the beam per unit length. If the beam has a constant cross
section and uniform mass distribution, rn = constant and Eq. (A.3.1) becomes
the equation of free vibration

EZa4Y T a 2 Y I J2Y -
(A.3.3)
m ax4 m ax2 at2

Substituting into (A.3.3) an assumed solution with separable variables y =

X(x)U(t), we arrive at two equations with single variables:

0-- --02
- (A.3.4a)
U
T -X "
_ - EIX'"
-- - - o2
(A.3.4b)
m X m X

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

41 8 Appendix 3

The first equation is a standard equation describing vibratory motion, the second
one describes the modal pattern. It can be rewritten as

XI\ - (yX’’ - k4X = 0 (A.3.4~)

where

The solution of Eq. (A.3.4~)can be generally expressed as (e.g., [2])

X = C, sinh r I x + Cz cosh r I x + Ci sin r2x + C4 cos r2x (A.3.5)

where

(A.3.6)

For the case of Fig. 7.30a of a double-supported beam, boundary conditions for
x = 0 and x = Z are X = 0 and X” = 0. Thus, C2 = C1= 0 and

C, sinh r,Z + C3 sin r2Z = 0; C l r ; sinh r,Z - C3r; sin r2Z = 0 (A.3.7)

Eq. (A.3.7) has nontrivial solutions if its determinant is zero or

(r: + r;)sinh rlZ sin r2Z = 0 (A.3.8a)

For any magnitude of axial force T = 0, rl Z > 0 and r2Z > 0. As a result, always
sinh rlZ > 0 and Eq. (A.3.8a) is equivalent to

sin r21 = 0 (A.3.8b)

Since r2Z = 0, Eq. (A.3.8b) is satisfied when

r2Z = n n ( n = 1, 2, 3 , . . .) (A.3.9)

Accordingly, natural frequencies con can be determined from the expression

(A.3.10)

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

Influence of Axial Force on Beam Vibrations 41 9

Transforming expression (A.3.10), an explicit expression for a, can be written

as

(A.3.1 la)

or

(A.3.1 lb)

where a,, = natural frequencies of the same beam without the axial load; T, =
Euler force for the beam; and n = order of the vibratory model. If T < 0 (com-
pressive force), then all a, are decreasing with increasing force, and a1becomes
zero at T,, when the beam buckles. If T > 0, then any increase in its magnitude
leads to a corresponding increase in all natural frequencies of the beam. This
effect is more pronounced for the lower natural frequencies, especially al, due
to a moderating influence of the factor lln2.
Although Eq. (A.3.1 la) was derived for a double-supported beam, the ge-
neric expression (A.3.11b) seems to be valid for other supporting conditions as
well.

REFERENCES

1. Rivin, E.I., Mechanical Design of Robots, McGraw-Hill, New York, 1988.

2. Craig, R.R., Structural Dynamics, John Wiley & Sons, New York, 1981.

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.